I wanted to thank everyone again for the assistance and advice delivered on
thread "[NSSound isPlaying] fails to indicate sound termination in Lion". I
know what to do with that one now, and if Apple ever says anything about my bug
report, I will let those of you who have helped most know what they said.
There is a related design issue, however, for which I no longer have a good
Cocoa solution; perhaps some of you could advise me.
The issue is synching animations -- visual ones -- with sounds. Think of it as
lip-synching. You will recall that what I wanted to do was have a cat-face
appear, open its mouth, say (play sound) "meow", then close its mouth and go
away. For this little effect to look best, the open-mouth animation should end
just as the sound begins, and the close-mouth animation should start just as
the sound ends. I know I can't count on that happening if I the user has too
many processes running and the underlying Unix scheduler decides to swap at
just the wrong time, but let's assume that my App is getting plenty of cycles.
There is no problem issuing [meow play] as the next instruction after the end
of the open-mouth animation; the issue is doing the best job of making the
close-mouth animation start just as the sound ends. The problem with a
call-back method, like the NSSound delegate method "didFinishPlaying:" is that
that method evidently works by posting an event of some sort to the main event
loop, or perhaps in a dispatch queue somewhere, and if there are other events
preceding it in either the event loop or the dispatch queue, the operation of
"didFinishPlaying:" may be delayed. (My app is a Lisp REPL, it occasionally
does some big chunks of work; fractional-second delays do happen, and a few of
those in the queue would noticeably mess up the timing of the animation.)
It was for that reason that I originally chose a polling loop
while( [meow isPlaying] )
;
to establish that the sound had ended. The code to close the cat's mouth
originally followed the end of the while loop. I did and do know that polling
loops are bad form for various reasons, but to get a cute and infrequently-used
special effect looking just right, I thought that a half-second polling loop
was a defensible coding choice.
That polling loop doesn't work under Lion -- so be it. And a call-back may be
delayed for the reasons just discussed.
The design issue is, what else can I do?
To give you a feel for an approach that would be faster, but that has other
disadvantages and might be difficult to implement in Cocoa, remember that back
when dinosaurs ruled the earth and we all knew how to program 8259s, we could
have just set up some kind of interrupt -- arranged that sound termination
cause code to run that preempts the processor's resources (or at least, the
main thread's), in order to run the close-mouth animation right NOW! I do know
how to use Unix's SIGINT facility, but I am not sure there is a way to get
sound termination to generate an interrupt, and I would be a bit uneasy having
an interrupt handler try to do something -- run the animation -- that normally
is handled in the main event loop.
So: In Cocoa, is there a reasonable approach to getting things to happen
quickly, that works when call-back on the main event loop isn't fast enough?
It is probably not a matter of great consequence for my situation, but it might
matter more to others, and it is an interesting design challenge in any case.
-- Jay Reynolds Freeman
---------------------
[email protected]
http://JayReynoldsFreeman.com (personal web site)
_______________________________________________
Cocoa-dev mailing list ([email protected])
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
https://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to [email protected]
